Lipoxygenase directed anti-inflammatory and anti-cancerous secondary metabolites: ADMET-based screening, molecular docking and dynamics simulation

Lipoxygenases (LOXs), key enzymes involved in the biosynthesis of leukotrienes, are well known to participate in the inflammatory and immune responses. With the recent reports of involvement of 5-LOX (one of the isozymes of LOX in human) in cancer, there is a need to find out selective inhibitors of 5-LOX for their therapeutic application. In the present study, plant-derived 300 anti-inflammatory and anti-cancerous secondary metabolites (100 each of alkaloids, flavonoids and terpenoids) have been screened for their pharmacokinetic properties and subsequently docked for identification of potent inhibitors of 5-LOX. Pharmacokinetic analyses revealed that only 18 alkaloids, 26 flavonoids, and 9 terpenoids were found to fulfill all the absorption, distribution, metabolism, excretion, and toxicity descriptors as well as those of Lipinski’s Rule of Five. Docking analyses of pharmacokinetically screened metabolites and their comparison with a known inhibitor (drug), namely zileuton revealed that only three alkaloids, six flavonoids and three terpenoids were found to dock successfully with 5-LOX with the flavonoid, velutin being the most potent inhibitor among all. The results of the docking analyses were further validated by performing molecular dynamics simulation and binding energy calculations for the complexes of 5-LOX with velutin, galangin, chrysin (in order of LibDock scores), and zileuton. The data revealed stabilization of all the complexes within 15 ns of simulation with velutin complex exhibiting least root-mean-square deviation value (.285 ± .007 nm) as well as least binding energy (ΔGbind = −203.169 kJ/mol) as compared to others during the stabilization phase of simulation.

The therapeutic use of natural products belongs to the oldest medical practices. According to one of the World Health Organization Estimates, almost 80% of the world's population utilizes plant products as drugs, especially in the developing countries (Wang, Calway, & Yuan, 2012). Globally, more than 85,000 plant species have been documented for their medicinal uses. Plant products being natural, are often considered as safe, though some side effects have also been reported for few plant products (Tavakoli, Miar, Zadehzare, & Akbari, 2012). Thus, for therapeutic applications, there is a need for more specific inhibitors of selected cancer targets coming from natural sources which are relatively safe with little or no side effects and also having a lower cost (Lahlou, 2013).
Alkaloids, flavonoids, and terpenoids constitute three major classes of plant-derived secondary metabolites which hold great promise for discovery and development of new pharmaceuticals for diverse human ailments including cancer. Importance of plant-derived pharmaceuticals in cancer treatment can be visualized by the fact that almost two-third of the anti-cancerous drugs available in the market are derived from plant sources (Qurishi, Hamid, Zargar, Singh, & Ajit, 2010;Safarzadeh, Shotorbani, & Baradaran, 2014). Furthermore, among the US Food and Drug Administration (FDA) approved anti-cancerous drugs during the period of 1984-1994, 60% were those isolated from natural sources specially plants (Kumar, George, Suresh, & Kumar, 2012). Among 65 new drugs registered for cancer treatment during the period 1981-2002, 48 drugs were obtained from natural products (Wang et al., 2012). In this direction, in the present paper, plant-derived 300 secondary metabolites (100 each of alkaloids, flavonoids, and terpenoids), having anti-inflammatory as well as anti-cancerous properties, have been screened for their pharmacokinetic properties and subsequently docked for identification of potent inhibitors of 5-LOX. The results of the docking analyses were further validated by performing molecular dynamics (MD) simulation and binding energy calculations for the 5-LOX complexes with promising secondary metabolites and compared with that of a well-known inhibitor (drug) of 5-LOX.

Preparation of protein structure
The crystal structure of human 5-LOX (PDB ID: 3O8Y) was retrieved from Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank (http://www. rcsb.org/pdb/home/home.do) (Berman et al., 2000). RCSB is the single, global archive for information pertaining to the 3D structure of macromolecules and their complexes, as determined by X-ray crystallography, NMR spectroscopy, and cryoelectron microscopy. The retrieved 5-LOX structure was cleaned and the hetero atoms (HETATM) of the receptor were removed and necessary changes were made using Discovery Studio version 3.1 (DS) (Accelrys Software Inc, 2013).

Preparation of ligands structure
Based on available literature and databases of plant secondary metabolites, namely, Naturally Occurring Plantbased Anti-cancer Compound-Activity-Target database (NPACT) (Mangal, Sagar, Singh, Raghava, & Agarwal, 2013) and database for Benzylisoquinoline Alkaloids (BIAdb) (Singla, Sharma, Kaur, Panwar, & Raghava, 2010), a total of 300 anti-inflammatory and anti-cancerous secondary metabolites were selected (Supplementary Tables S1-S3). Commonly used drug, zileuton was taken as reference molecule (ligand) for the docking as well as simulation analyses at the active site of 5-LOX. The mol files of these inhibitors were retrieved from the ChemSpider (http://www.chemspider.com/) and Pub-Chem (http://pubchem.ncbi.nlm.nih.gov/) databases. Furthermore, the mol files of ligands were visualized and the energy minimization was done through 'in situ ligand minimization' module of DS.

Pharmacokinetic analyses of secondary metabolites
2.3.1. ADMET analyses Absorption, distribution, metabolism, excretion, and toxicity (ADMET) analyses of 300 selected secondary metabolites (100 each of alkaloids, terpenoids, and flavonoids) were performed using DS. ADMET analyses of potential drug candidates describe the disposition of these compounds within an organism (Balani, Miwa, Gan, Wu, & Lee, 2005;Hou & Xu, 2004). The longterm success of a drug is determined not only by good efficacy, but also on having acceptable ADMET descriptors (Segall, Beresford, Gola, Hawksley, & Tarbit, 2006). Toxicity predictions of the selected secondary metabolites were done using weight of evidence (WOE) parameter of carcinogenicity of TOPKAT module of DS. WOE predictions are in accordance with animal model guidelines of FDA and National Toxicology Program (NTP), USA.

Drug likeness
Analysis of drug likeness of the ADMET screened metabolites was done using DS, by assessing four properties, namely H-bond donor, H-bond acceptor, molecular weight, and AlogP (partition coefficient) (Lipinski, Lombardo, Dominy, & Feeney, 2001).

Molecular docking studies
Docking analyses of 5-LOX with the pharmacokinetically screened metabolites were performed using 'LibDock module' (Rao, Head, Kulkarni, & LaLonde, 2007) of DS. LibDock is a high throughput docking algorithm that positions catalyst generated ligand conformations in the protein active site, based on polar interaction sites (hotspots). LOX active site was predicted, based on the PDB site record, using 'Define and Edit Binding Site module' of Receptor-Ligand Interactions tool of DS. A binding sphere (radius) of 10.3 Å with a grid dimensions of 4.967 × 21.402 × .272 Å was fixed in LibDock, to define the binding region on protein, where receptor-ligand interactions were allowed to occur. The CHARMm Force Field was used for energy minimization of the ligands. CAESAR (Conformer Algorithm based on Energy Screening and Recursive build-up) was used for generating conformations with surface grid step value of 18 and maximum number of runs as 255. The methodology was validated by performing re-docking by varying parameters. For comparison, a known inhibitor (drug) of 5-LOX, namely zileuton was used as a reference compound for docking analyses. To better understand the interactions between protein residues and bound ligands, the 3D diagrams were generated that also helped to identify the binding site residues.

MD simulation
Interactions of 5-LOX with the top ranked secondary metabolites (based on LibDock score) were investigated and compared with that of drug (zileuton) through 25 ns of MD simulation using GROMACS 4.5.5 package (Pronk et al., 2013) with GROMOS96 43a1 force field. The initial conformations were taken from the docking results. The topology of the protein was generated using the GROMACS program, while the topology of the ligand molecules were built using PRODRG server (http://davapc1.bioch.dundee.ac.uk/cgi-bin/prodrg). The system was immersed in a water filled cubic box of 1 Å spacing containing 36,705 water molecules using extended simple point charge (SPC), a three-point water model with periodic boundary conditions. The total charge of the solvated system was neutralized by adding counter ions to the system. To release conflicting contacts, energy minimization was performed sequentially using 500 steps of steepest-descent minimization with and without positions restraints for the heavy atoms. Also, 1500 steps of conjugated gradient minimization with a convergence criterion of .001 kcal/ (mol × Å) were applied to the solvent (water) molecules and the entire model system, respectively. After energy minimization, the entire system was gradually heated from 0 to 300 K and was submitted to 100 ps of equilibration using the NVT (constant volume and temperature) ensemble with position restrained by a weak harmonic potential for the heavy atoms of the protein complexed with ligands followed by the 100 ps equilibration in the NPT (constant pressure and temperature) ensemble. Finally, the full system was subjected to 25 ns MD simulation at 300 K temperature and 1 bar pressure without restraints. The periodic boundary condition was used and the motion equations were integrated by applying the leap-frog algorithm with a time step of 2 femtoseconds (Gonçalves, França, Figueroa-Villar, & Pascutti, 2011).

Binding free-energy analyses
The binding free energy of the complexes of 5-LOX with top ranked secondary metabolites and drug, during the last stable 10 ns period of MD simulation analyses, were computed using g_mmpbsa tool of GROMACS (Kumari, Kumar, Open Source Drug Discovery Consortium, & Lynn, 2014), based on the molecular mechanics/Poisson-Boltzman surface area (MM/PBSA) method (Kollman et al., 2000). The binding energy calculations were performed for 200 snapshots taken at an interval of 50 ps during the last stable 10 ns period of MD trajectory.

ADMET-based screening
In view of failure of many FDA-approved drugs at later stages, due to undesirable pharmacokinetic and toxicity properties, there is a need to investigate inhibitors for therapeutic applications having long-term success, based on the parameter of ADMET criteria (Hop et al., 2008;Oprea, 2002;Ponnan et al., 2013). Thus, 300 potential anti-inflammatory and anti-cancerous secondary metabolites, belonging to 100 each of alkaloids, flavonoids, and terpenoids from plant sources, were selected from relevant literature and databases (NPACT, BiaDB) and were subjected to ADMET screening in a step wise manner as described in following sections.

Human intestinal absorption and blood brain barrier
Three hundred selected anti-inflammatory and anticancerous secondary metabolites were subjected to screening for human intestinal absorption (HIA) and blood brain barrier (BBB) descriptors of ADMET. Results are presented in Figure 2. HIA levels of 0, 1, 2, and 3 reflect absorption levels to be good, moderate, poor, and very poor, respectively. BBB levels of 0, 1, 2, 3, and 4 reflect penetration level to be very high, high, medium, low, and undefined, respectively. It is noteworthy that, out of 300 metabolites, only 62 alkaloids, 51 flavonoids, and 46 terpenoids were found to fulfill the set criteria (PSA < 140 Å 2 and AlogP98 < 5) (Egan, Merz, & Baldwin, 2000) at both 95% as well as 99% confidence limit ellipses. The area encompassed by the ellipse is a prediction of good absorption on the basis of ADMET prediction. Polar surface area (PSA) has been shown to be inversely related to HIA and cell membrane permeability ( Figure 2) (Palm, Stenberg, Luthman, & Artursson, 1997).

Plasma protein binding
Sixty-two alkaloids, fifty-one flavonoids, and forty-six terpenoids, selected on the basis of HIA and BBB descriptors, were subjected to screening for plasma protein binding (PPB) parameter of ADMET. Molecules which possess the Bayesian score of −2.226 or less, are predicted to be highly bound (≥90%) to plasma proteins and therefore do not fulfill (i.e. false) PPB criteria of ADMET. Thus, based on this criteria, 39 alkaloids, 36 flavonoids, and 31 terpenoids were found to have good PPB property as possessing Bayesian score of more than −2.226 (i.e. true) thereby suggesting that they are not likely to be highly bound to carrier proteins in the blood.

Cytochrome P450 2D6 (CYP2D6) Binding
Thirty-nine alkaloids, thirty-six flavonoids, and thirty-one terpenoids, selected on the basis of PPB descriptor, were further subjected to screening for CYP2D6 binding criteria of ADMET. CYP2D6 is involved in the metabolism of a wide range of metabolites in the liver. Molecules having Bayesian score of <.162 are predicted as non-inhibitors of CYP2D6. Thus, based on this criteria, 29 alkaloids, 30 flavonoids, and all 31 terpenoids were found as non-inhibitors of CYP2D6 thereby successfully fulfilling this criteria.

Aqueous solubility
Twenty-nine alkaloids, thirty flavonoids, and thirty-one terpenoids, selected on the basis of CYP2D6 binding descriptor, were subjected to screening for solubility criteria of ADMET. Solubility levels of 0, 1, 2, 3, 4, and 5 reflect absorption levels to be extremely low; no, very low, but possible; yes, low; yes, good; yes, optimal and no, too soluble, respectively. Based on the set parameter of solubility, only 24 alkaloids, all 30 flavonoids, and 30 terpenoids were found to possess low to optimal aqueous solubility thereby successfully fulfilling this criteria.

Toxicity
Twenty-four alkaloids, thirty each of flavonoids, and terpenoids, selected on the basis of solubility descriptor, were subjected to screening for toxicity criteria of ADMET. For determining toxicity, the WOE parameter of carcinogenicity of TOPKAT module of DS was considered. WOE prediction is based on the rodent carcinogenicity model derived from data provided by the FDA-Center for Drug Evaluation and Research (CDER) and from uniform studies selected after critical review of technical reports on rodent carcinogenicity studies conducted by National Cancer Institute (NCI) and NTP which compute the probability of a submitted chemical structure being a carcinogen. Thus, based on WOE prediction parameter, 18 alkaloids, 26 flavonoids, and 9 terpenoids were found to be non-carcinogenic (NC). Results of ADMET analyses of finally screened 53 metabolites viz. 18 alkaloids, 26 flavonoids, and 9 terpenoids showing various ADMET descriptors, are presented in Table 1.

Drug likeness of ADMET screened metabolites
The drug-like properties of the finally ADMET screened 53 metabolites viz. 18 alkaloids, 26 flavonoids, and 9 terpenoids were analyzed on the basis of Lipinski's rule of five. Results are presented in Table 2. Data revealed that all the successfully ADMET screened secondary metabolites successfully conceded the Lipinski's rule of five.

Molecular docking analyses of the pharmacokinetically screened secondary metabolites with 5-LOX and its comparison with the drug zileuton
A total of 53 pharmacokinetically screened metabolites (fulfilling the ADMET and Lipinski's rule of five criteria) belonging to 18 alkaloids, 26 flavonoids, and 9 terpenoids, were compared for their efficacy of inhibition of 5-LOX using molecular docking approach. For comparison, a known inhibitor (drug) of 5-LOX, namely zileuton, which has also been shown to suppress cancer progression, was analyzed (Berger et al., 2007;Guo & Nie, 2012). Results of docking analyses are presented in Table 3. Thus, only three alkaloids, six flavonoids, and three terpenoids were found to dock successfully with 5-LOX. It is noteworthy that flavonoid velutin (a flavone) was found to be the most potent inhibitor of 5-LOX among all the docked metabolites as well as to that of drug zileuton.
To the best of our knowledge, this is the first report where velutin has been shown to directly inhibit 5-LOX. However, velutin has been shown to effectively inhibit the expression of proinflammatory cytokines, tumor necrosis factor-α, and interleukin (IL-6) at low micromolar concentrations via inhibiting nuclear factor (NF)-κB activation and p38 and c-Jun N-terminal kinase (JNK) phosphorylation (Kang et al., 2011;Xie et al., 2012) thereby suggesting its role as an anticancerours compound. Velutin has also been reported to exhibit in vitro cytotoxicity against human nasopharynx carcinoma (KB) cells (IC 50 = 4.8 μM) (Zahir et al., 1996). Flavonoids baicalin and catechin from Acacia catechu have been reported to inhibit 5-LOX in lipopolysaccharide stimulated peritoneal rat macrophages (Altavilla et al., 2009).
3.4. Active site mapping of 5-LOX with regards to residues interacting with most potent flavonoid, velutin and its comparison with the drug zileuton Interactions of the most potent flavonoid, velutin and the drug, zileuton with 5-LOX, at the active site, were analyzed and compared. Data are shown in Table 4 and Figure 3. Thus, from the data, it is evident that velutin and zileuton both interacted with 5-LOX in similar manner as majority of interacting residues, namely Phe177, Gln363, Thr364, His367, Leu368, His372, Ile406, Asn407, Ala410, Leu414, Ile415, Phe421, His432, Leu607, and Ile673 were found to be common. Furthermore, binding of zileuton to 5-LOX was found to involve one H-bond interaction (H19-Gln363: O at a distance of 1.870 Å) while, binding of velutin to 5-LOX did not involve any H-bond interaction at the active site. In conformity with our results, Eren, Macchiarulo, and Banoglu (2012) have reported the involvement of almost similar interacting residues of 5-LOX complexed with a number of 5-LOX inhibitors. Furthermore, Leu368, Leu373, Ile406, Leu414, and Leu607 (forming a region of conserved hydrophobic side chains), accommodating the pentadiene moiety of the substrate for catalysis, have been reported to be conserved in all the AA metabolizing LOXs (Gilbert et al., 2011).

MD simulation analyses
The stability of the docked complexes of 5-LOX with each of the top 3 ranked flavonoids, namely velutin, galangin, chrysin were analyzed and compared with that of the drug zileuton using 25 ns MD simulation analyses.
For comparison, the 5-LOX Apo as well as those of the ligands alone, namely velutin, galangin, chrysin, and zileuton (drug) was also analyzed using 25 ns MD simulation. Results are shown in Table 5 and Figure 4. Thus, it may be suggested that all the 5-LOX complexes with Furthermore, to probe the flexibility of the various complexes as well as 5-LOX Apo, the root-mean-square  fluctuation (RMSF) for Cα atoms of all the residues were compared for 25 ns MD simulations ( Figure 5). It is noteworthy that the RMSF trajectories of all 5-LOX complexes as well as 5-LOX Apo demonstrated similar backbone fluctuations suggesting stability of the complexes with a relatively more rigid and stable structure for complex of velutin.

Binding free-energy analyses
In order to have deeper insight into the stability of the 5-LOX complexes with those of the velutin, galangin, chrysin, and zileuton (drug), the binding free energies of these complexes were computed using MM/PBSA method during the last stable 10 ns of MD simulation. Results are presented in Table 6. From the data presented  in Table 6, it is evident that the complex of 5-LOX with velutin exhibited a binding energy (ΔG bind ) of −203.169 kJ/mol, which is significantly lesser than those of complexes of 5-LOX with galangin, chrysin, and zileuton (drug). Based on these observations, it may be suggested that the stability of 5-LOX-velutin complex was significantly higher than those of others. Overall, MM/PBSA binding free-energy analyses corroborated with the results of molecular docking and dynamics simulation analyses.

Conclusion
The present paper describes the results of screening and identification of potent anti-cancerous plant-derived secondary metabolites, targeted toward a key enzymatic anti-inflammatory target, namely 5-LOX (one of the isozymes of LOX in human). Pharmacokinetic screening of 300 plant-derived anti-inflammatory and anti-cancerous secondary metabolites (100 each of alkaloids, flavonoids, and terpenoids) revealed only 18 alkaloids, 26 flavonoids, and 9 terpenoids as possible candidate drug molecules. The potency and efficacy of the successfully screened metabolites were further analyzed by performing molecular docking analyses which revealed that only three each of alkaloids and terpenoids, while six of flavonoids were docked successfully with 5-LOX. Among them, flavonoids velutin, galangin, chrysin, baicalein, and genkwanin were ranked higher as compared to alkaloids and terpenoids with velutin as the most potent inhibitor. The results of MD simulation and binding energy (ΔG bind ) calculations revealed least RMSD (.285 ± .007 nm) and ΔG bind (−203.169 kJ/mol) values for LOX-velutin complex as compared to those of other complexes suggesting velutin as the most potent inhibitor of 5-LOX. Thus, the present work makes a foundation  for the development of velutin as potent anti-cancerous drug acting through inhibition of an inflammatory enzymatic target 5-LOX.

Supplementary material
The supplementary material for this paper is available online at http://dx.doi.org/10.1080/07391102.2016. 1159985.